Simply put, the numbers are grossly exaggerated.
To fully prove one way or another, you would need to model a sample population over a long period of time and see how the male/female split diverges. I have seen this done (by someone who was trying to illustrate this theory for me), but to go over it here would be next to impossible. The key is that when modeling, your results reflect the inputs or assumptions that you set initially. It is easy to alter the assumptions to tailor the solution.
I felt strongly that the model inputs were inappropriate. When I asked the person to use my suggestions as inputs instead, there was almost no divergence.
Has anyone else here done modeling of this theory, or has everone just accepted it because it sounded nice? You know what else sounds nice? The sun going around the Earth, because then we are the center of the universe.